revLISP (actually is scheme...) Some doubts on implementation.

[Dar Scott]

On Thursday, March 25, 2004, at 07:36 PM, Andre Garzia wrote:

> Scheme is a LISP-like system. The interpreter is based on cells. Cells 
> have some attributes like pointers (sorry, this is C-like talk, I 
> should have said references) to other cells and some info on it's 
> container and ID... How should I implement a structure in Rev, should 
> I go for keyed arrays? can an array contain another array as an 
> element?

Currently, an array cannot contain another array as an element.  There 
is a request for that, since that can be interpreted as a 
multidimensional array.  I hope it comes.  This might also allow the 
use of arrays as property values.

There are a couple ways to go for a generic simple lisp  depending on 
how much you want to do.  For Scheme you may need to modify this.

1
If you do not share or mutate shared data, then a cons tree or list can 
be a single string structured in a way that has pairs.  For example, 
suppose you use the first letter for type info.  A "C" means cons and 
other letters refer to other types.  After the C is 9 digits.  This is 
the length of the CAR or first of the pair.  The data after the first 
is the rest or the CDR.  These can be nested arbitrarily deep.  
Philosophically, this is much like how I made "boxes".

2
Another way to go is to use a global array (or several) as your memory 
space.

A.  One array
Each element is a leaf value or a cons cell.  The first letter 
designates the type.  A cons cell after the type is a pair of indexes 
into the array.  You can probably come up with a better way to organize 
this.

B.  Three arrays
Like above but one array for the type, one for the value and one for 
the second value if the type is a cons.

You can add reference counting or garbage collection to either of A or 
B.  You can share sub structures and you can mutate parts, even shared 
parts.

Your exercise might help you think up all kinds of ways to store info 
as Revolution values.  I don't know how XML data is stored; it might 
give a clue to a third method.

Dar Scott